PROBLEM with SPINE: Setting Poller interval to 1 Minute

[yaun]

Hello everybody.
Please help!
I have cacti 0.8.7g with spine, PA and some plugins.
No graphs, no data sources, yet, but 222 hosts monitored thru ICMP and/or SNMP.
I have poller interval set to 5 minutes and cron interval to 5 minutes, too.
Everything works fine.
The fact is that 5 minutes is too much for me, I'd like to have 1 minute polling time.
So I set Poller Interval to 1 minute. Cron to 5 as usual.
Two errors recurring:
POLLER: Poller[0] Maximum runtime of 298 seconds exceeded. Exiting.
AND
SPINE: Poller[0] ERROR: Spine Timed Out While Processing Hosts Internal

Hosts aren't monitored anymore.
If I set back poller interval to 5 minutes, no errors, everything works good, even email notification.
Poller to 1 minute: crash, nothing works
Poller to 5 minute: OK
I didn't change anything else. Everything is set to default.
I've tried to change Maximum Concurrent Poller Processes, Maximum Threads per Process, no result.
I've rebooted the server, no way.

[gandalf]

Please see http://forums.cacti.net/about23885.html
R.

[yaun]

Well, I read it before posting.
You seem to be confident that there's an answer in that post.
I'll read it another time and I'll let you know.
However, I'm a bit pessimistic.

[gandalf]

Yep, I know that this post is lengthy. But IMHO it is the best one covering that topic. You will find out, that RRA definitions are important for correct processing
R.

[yaun]

Ok. I've read it for the second time. Very interesting.
I didn't think that was my problem, since I hadn't graphs at all.
I was only polling host availability (I'd do graphs when I set everything up and working).
However, I've changed RRAs according to the post.
Then I changed poller interval to 1 minute.
Same errors in the log.
I've disabled everything: disabled data templates, I've no data sources, I've removed all rrd files generated when I'd launched auto-discovery (entering manually 221 hosts, would be a little too much).
I've checked cron.d/cacti, crontab and every other cron file even if I didn't even open them.
Only one entry in cron.d/cacti, 5 minute interval.
Nada!
So, with my cacti fresh installation gone to hell, every parameter changed, I tried to go back to 5 minute poller interval.
It shouldn't work anymore, isn't it? I've changed RRA and so on...
Actually, It returned to work. I wouldn't ever been able to a graph again without losing 1 day in debugging, but, spine and cmd.php work like a charm.
I'm a little discouraged. I really can't see what I'm doing wrong!
I really need something like cacti, it's the NMS that more fits my needs. But making it work is really a pain in the...

[gandalf]

You are aware, that changing RRA settings won't propagate to existing rrd files?
And to be honest, I do not really understand what you're doing. You told us, that you're only using ICMP host alive detection. And then, you're using graphs (which?). I can't make a story out of this
R.

[yaun]

I'm simply trying to monitor if my 222 hosts are up or down, polling with interval of 1 minute.
Mind that I'm using ICMP only if the device monitored has no SNMP.
I HAD graphs but they were auto-generated by the discovery plugin, which I used to populate cacti.
I've deleted all graphs and all data sources BEFORE attempting to change poller interval, and BEFORE changing RRA.
When I've changed poller interval I had only 222 devices, no graphs at all.
If I succeed in changing poller interval to 1 minute, I will consider creating graphs.

In this scenario, with 5 minute interval, everything works.
When I change it to 1 minute, the poller (SPINE or CMD.PHP) fails.
Changing RRA settings didn't do anything at all.

I hope that the everything is clear now, sorry if it wasn't.

[yaun]

How to have the problem:
1. install new Debian 6.0
2. install cacti 0.8.7g thru apt-get
3. install cacti-spine thru apt-get
4. install PA thru patch
5. Change poller interval.

I've tried on another server, same result.

[wardrobe]

Im getting the same error on a new debian 6 box, through a debian package install

[wardrobe]

manually changing the poller from 5 to 1 and editing the cron job 5 to 1 seems to of resolved it